CI: Add auto-deployment via dokku.
[akkoma] / lib / pleroma / web / rich_media / parser.ex
1 # Pleroma: A lightweight social networking server
2 # Copyright © 2017-2019 Pleroma Authors <https://pleroma.social/>
3 # SPDX-License-Identifier: AGPL-3.0-only
4
5 defmodule Pleroma.Web.RichMedia.Parser do
6 @parsers [
7 Pleroma.Web.RichMedia.Parsers.OGP,
8 Pleroma.Web.RichMedia.Parsers.TwitterCard,
9 Pleroma.Web.RichMedia.Parsers.OEmbed
10 ]
11
12 @hackney_options [
13 pool: :media,
14 recv_timeout: 2_000,
15 max_body: 2_000_000,
16 with_body: true
17 ]
18
19 def parse(nil), do: {:error, "No URL provided"}
20
21 if Mix.env() == :test do
22 def parse(url), do: parse_url(url)
23 else
24 def parse(url) do
25 try do
26 Cachex.fetch!(:rich_media_cache, url, fn _ ->
27 {:commit, parse_url(url)}
28 end)
29 rescue
30 e ->
31 {:error, "Cachex error: #{inspect(e)}"}
32 end
33 end
34 end
35
36 defp parse_url(url) do
37 try do
38 {:ok, %Tesla.Env{body: html}} = Pleroma.HTTP.get(url, [], adapter: @hackney_options)
39
40 html |> maybe_parse() |> clean_parsed_data() |> check_parsed_data()
41 rescue
42 e ->
43 {:error, "Parsing error: #{inspect(e)}"}
44 end
45 end
46
47 defp maybe_parse(html) do
48 Enum.reduce_while(@parsers, %{}, fn parser, acc ->
49 case parser.parse(html, acc) do
50 {:ok, data} -> {:halt, data}
51 {:error, _msg} -> {:cont, acc}
52 end
53 end)
54 end
55
56 defp check_parsed_data(%{title: title} = data) when is_binary(title) and byte_size(title) > 0 do
57 {:ok, data}
58 end
59
60 defp check_parsed_data(data) do
61 {:error, "Found metadata was invalid or incomplete: #{inspect(data)}"}
62 end
63
64 defp clean_parsed_data(data) do
65 data
66 |> Enum.reject(fn {key, val} ->
67 with {:ok, _} <- Jason.encode(%{key => val}) do
68 false
69 else
70 _ -> true
71 end
72 end)
73 |> Map.new()
74 end
75 end